F4 F4 F4 F4 F4 F4 34 41 | . . . . . . 4A
63 71 20 34 00 F4 F4 F4 |cq 4 . . . .
F4 F4 F4 F4 34 41 63 71 |. . . .4Acq
20 35 00 F4 F4 F4 F4 F4 |5 . . . . . . .
F4 4F 34 41 63 71 20 36 |. . 4Acq 6
Any data in the NVRAM was lost. I had pulled the battery packs off the original RAM to get it to boot. To the left of the vertical bar is the hex. To the right is the ASCII . that turned out to be a valuable clue.
From looking at the names and the diagnostic manual, I'm pretty sure it's storing the error codes for each of the subsystems. If you start dumping memory from 0x1DD70 in 0x60 byte blocks you can read the names of the routines in the diagnostic manual.
I'll also see if I can come up with a way to step the aliased ramp through all the aliases. But so far I've not seen any evidence of systematic timebase error except when it gets borked and reports ~320 ps for the calibrator or SD-24 rise time measurement.
BTW I'd had concerns about screen jitter, but that turned out to be a loose IEC connector. I had not pushed it in quite all the way.